Antifouling nanofiltration membranes for membrane bioreactors from self-assembling graft copolymers

Journal of Membrane Science 285 (2006) 81–89

Antifouling nanofiltration membranes for membrane bioreactors from self-assembling graft copolymers Ayse Asatekin a , Adrienne Menniti b , Seoktae Kang d , Menachem Elimelech d , Eberhard Morgenroth b,c , Anne M. Mayes a,∗ b

a Massachusetts Institute of Technology, Department of Material Science and Technology, Cambridge, MA 02139, United States Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States c Department of Animal Sciences, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States d Department of Chemical Engineering, Yale University, New Haven, CT 06520, United States

Received 13 May 2006; received in revised form 24 July 2006; accepted 30 July 2006 Available online 7 August 2006

Abstract Ultrafiltration (UF) membranes, often employed in membrane bioreactors (MBRs), exhibit high susceptibility to fouling by extracellular polymeric substances (EPS). As potential alternatives, commercial polyvinylidene fluoride (PVDF) UF membranes were coated with the amphiphilic graft copolymer poly(vinylidene fluoride)-graft-poly(oxyethylene) methacrylate, PVDF-g-POEM, to create thin film composite (TFC) nanofiltration membranes. Pure water permeabilities up to 56 L/m2 h MPa were obtained at pressures of 0.21 MPa (30 psi). The new TFC NF membranes exhibited no irreversible fouling in 10-day dead-end filtration studies of model organic foulants bovine serum albumin, sodium alginate and humic acid at concentrations of 1000 mg/L and above. Dead-end filtration of activated sludge from an MBR (1750 mg/L volatile suspended solids, VSS) resulted in constant flux throughout the 16 h filtration period. Fouling performance of the TFC NF membrane and effluent water quality were substantially improved in all cases over that for the base PVDF UF membrane. Utilizing the atomic force microscope (AFM) colloid probe technique, the measured interaction force profiles indicated the presence of repulsive steric interactions, which likely prevent the attachment of foulants to the TFC NF membrane. Similarly, the adhesion (pull-off) curves reveal the absence of foulant adhesion to the TFC NF membrane surface, even in the presence of divalent calcium ions. In contrast, when such force measurements are carried out with the base PVDF UF membrane, substantial adhesion forces are registered. © 2006 Elsevier B.V. All rights reserved. Keywords: Ultrafiltration; Nanofiltration; Membrane bioreactor; Fouling; Extracellular polymeric substances

1. Introduction Membrane bioreactors (MBR) – systems that combine conventional biological wastewater treatment using suspended biomass with membrane separation – are an attractive alternative to conventional activated sludge treatment using secondary sedimentation. MBRs offer the advantages of higher product water quality and reduced footprint [1]. However, the widespread application of MBRs is constrained by membrane fouling [1,2]. In a survey of papers published between 1991 and 2004, more than one third of studies on fundamental aspects of MBRs were related to membrane fouling [3].

∗

Corresponding author. Tel.: +1 617 253 3318; fax: +1 617 452 3432. E-mail address: [email protected] (A.M. Mayes).

0376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2006.07.042

A variety of constituents in water can lead to membrane fouling, including dissolved inorganic or organic compounds, colloids, bacteria and suspended solids [4]. Biofouling is largely attributable to accumulated extracellular materials, rather than individual bacterial cells or microbial flocs [5–7]. These extracellular materials, including soluble microbial products (SMP) and extracellular polymeric substances (EPS), consist mainly of polysaccharides, proteins, and natural organic matter (NOM) [8–12]. The present work thus focuses on addressing fouling from these three classes of biomolecules and by activated sludge from an MBR. Membranes employed in MBRs are typically porous ultrafiltration (UF) membranes. For UF membranes, improvement in fouling resistance is largely achieved by surface graft polymerization of hydrophilic monomers [13]. This method has the disadvantage of employing high-energy methods, such as

82

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

␥-irradiation or plasma treatments, resulting in a significant increase in membrane fabrication cost. Undesirable side reactions include the polymerization of ungrafted chains, which can leach from the surface during use. Such graft-polymerized layers can also block surface pores and reduce the intrinsic membrane permeability, while leaving the membrane susceptible to internal pore fouling. To improve effluent quality while substantially eliminating internal pore fouling, nanofiltration (NF) membranes offer a potential alternative to UF membranes for MBRs [14–16]. NF membranes have properties between those of reverse osmosis (RO) and UF. They combine the advantage of low operational pressures with size cut-off on the molecular scale [4]. NF is used in wastewater treatment, purification of ground and surface waters to produce drinking water, and for reverse osmosis pretreatment [4]. It also has applications in many industries, such as the textile, pharmaceutical, food, and pulp and paper industries [17,18]. NF MBR systems, unlike UF-based ones, have the potential for rejecting low molecular weight contaminants such as endocrine disrupting chemicals, pharmaceuticals, and pesticides that can be hazardous to human health [16]. In their study of NF MBRs, Choi et al. [14,15] noted that high flux NF membranes with high organic matter rejection and low salts rejection are needed to improve the practicability of such systems. Because of the high rejection of organics in NF MBR systems, they can further reduce fouling of RO membranes when used in advanced wastewater reclamation in place of UF MBRs. Commercial NF membranes are generally thin film composites (TFCs) consisting of a charged polyamide interfacially polymerized onto a polysulfone UF membrane support [17,19]. The dense nature of the polyamide layer serves to preclude internal pore fouling, but also dramatically reduces flux. Moreover, the chemical nature of this interfacial layer leaves NF membranes susceptible to fouling by solvated or suspended charged species in feed waters. Recently, we developed a new thin film composite NF membrane comprised of a commercial polyvinylidene fluoride (PVDF) UF membrane coated with the amphiphilic graft copolymer poly(vinylidene fluoride)-graft-poly(oxyethylene) methacrylate, PVDF-g-POEM [20,21]. Microphase separation of the PVDF backbone and short polyethylene oxide (PEO) side chains results in an interconnected network of hydrophilic, charge-neutral “nanochannels” ∼2 nm in width that allow the passage of water and small molecules through the coating. In preliminary investigations, these membranes demonstrated the ability to fractionate like-charged molecular dyes and to completely resist fouling by model oil/water microemulsions in short-term dead-end filtration studies [20]. The high PEO content (>40 wt.%) of the PVDF-g-POEM coating should likewise lend resistance to fouling by proteins and other charged biomolecules [22,23]. In this study, the fouling behavior of PVDF-g-POEM/PVDF TFC NF membranes is investigated, first employing bovine serum albumin, sodium alginate, and humic acid as representatives of the three important classes of biomolecule foulants in MBRs: proteins, polysaccharides and NOM, respectively

[10–12]. Fouling behavior is characterized in 10-day filtration studies using 1000 mg/L feed solutions. Filtration studies using activated sludge from an aerobic MBR as the feed solution are also described. The PVDF-g-POEM coated membranes are shown to completely resist irreversible fouling, defined as fouling that cannot be recovered by a pure water rinse, while providing substantially improved effluent quality over the PVDF UF base membrane control. Atomic force microscope (AFM) colloid probe measurements confirm the absence of organic adhesion to the PVDF-g-POEM membrane and further suggest that the fouling resistance is attributable to a long-range steric repulsive force between the foulants and the membrane. 2. Experimental 2.1. Materials Poly(vinylidene fluoride) (PVDF, Mn ∼ 107 kg/mol), poly(ethylene glycol) methyl ether methacrylate (POEM, Mn = 475 g/mol), N-methyl pyrrolidone (NMP), 4-methoxyphenol (MEHQ), N,N,N ,N ,N -pentamethyldiethylenetriamine (PMDETA), bovine serum albumin (BSA, 66.5 kDa), and phosphate buffer saline (PBS) packets were purchased from Sigma–Aldrich (St. Louis, MO). Copper(I) chloride (CuCl), basic-activated alumina, poly(ethylene glycol) (PEG, Mn = 600 g/mol), N,Ndimethyl formamide (DMF), hexane, ethanol, tetrahydrofuran (THF), deuterated dimethyl sulfoxide (DMSO-d6 ), humic acid, calcium chloride (CaCl2 ) and sodium alginate were purchased from VWR (West Chester, PA). All chemicals and solvents were reagent grade, and were used as received. PVDF400 ultrafiltration membranes, purchased from SEPRO Inc. (Oceanside, CA) were used as the base membrane. Deionized water was produced from a Millipore Milli-Q unit. The molecular weights of sodium alginate and humic acid were measured by static light scattering (Wyatt MiniDawn). The experiments were performed using solutions in water at three different concentrations, and measurements at three different angles. The weight-average molecular weight of sodium alginate was found to be 29,000 g/mol, and that of humic acid was calculated to be 17,000,000 g/mol. The unusually high molecular weight measured for humic acid indicates the formation of aggregates in water [24]. The activated sludge used in this study was cultured within an aerobic membrane bioreactor having a porous liner manufactured into its walls through which fluid is passed by gravity [25]. The liner consists of filter grade porous polyethylene (Atlas Minerals & Chemicals, Mertztown, PA) with a thickness of 0.48 cm and a nominal pore size of approximately 25 ␮m. The reactor had a diameter of 9 in. (22.86 cm), a volume of 9.4 L and was configured into a standard reactor geometry according to Holland and Chapman [26]. Mixing was accomplished with a 3-in. (7.62 cm) diameter Rushton impellor at a speed of 150 revolutions per minute (rpm). The reactor was operated with a hydraulic retention time of 18 h and a solids retention time of 28 days. The influent consisted of 150 mg/L of ammonia nitrogen and acetate at a concentration of 350 mg/L chemical oxygen demand (COD) resulting in

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

a mixed liquor volatile suspended solids (MLVSS) concentration of approximately 1800 mg/L VSS. COD and MLVSS were measured according to [27]. Ammonia nitrogen was measured using a microtiter-based method [28]. 2.2. Synthesis of PVDF-g-POEM POEM was grafted to PVDF following a slightly modified version of the atom transfer radical polymerization (ATRP) approaches previously published [23,29]. PVDF (5 g) was dissolved in NMP (50 mL) in a conical flask at 50 ◦ C. The solution was cooled to room temperature and transferred to a 250 mL Schlenk flask. POEM (50 mL), CuCl (0.04 g), and PMDETA (0.26 mL) were added to the reaction vessel, which was subsequently sealed. Nitrogen gas was bubbled through the reaction mixture for 20 min. After a total reaction time of 1 h, MEHQ (2.5 g), dissolved in approximately 10 mL of THF, was added to the reaction mixture, which was then diluted with approximately 100 mL of THF. This mixture was passed through a column of basic-activated alumina and precipitated in a 10:3 mixture of hexane and ethanol. The product was redissolved and reprecipitated twice, then dried in a vacuum oven overnight. POEM content of the copolymer was determined by 1 H nuclear magnetic resonance (NMR) spectroscopy in deuterated DMSO using a Bruker DPX 400 spectrometer. 2.3. Preparation of coated membranes The coating solution was prepared by dissolving 1 g of PVDF-g-POEM copolymer and 1 g of PEG in 4 mL DMF at approximately 50 ◦ C. This solution was passed through a 1-␮m syringe filter (Whatman) and degassed under vacuum for approximately 1 h. Membranes were coated using a control coater (Testing Machines Inc., Ronkonkoma, NY). The PVDF400 base membrane was fixed onto the coater, and the coating bar (number 4, nominal film thickness 40 ␮m) was inserted. The coating solution was poured onto the base membrane to form a thin line about 0.5 cm from the coating bar, and the coater was used to move the bar at a constant reproducible speed (speed level 4 on the instrument). After 5 min, the membrane was immersed in a bath of ethylene glycol for 30 min, then rinsed in an isopropanol bath for 10 min and air dried. Membrane morphology was characterized using a JEOL 5910 scanning electron microscope (SEM) operating at 5 kV. The membranes were fractured in liquid nitrogen for crosssectional observation, and sputter coated with gold–palladium for SEM imaging. 2.4. Filtration experiments Circular pieces were cut from coated membranes and wetted in water for at least 1 h before performing filtration experiments. Long-term fouling experiments were performed on 25 mm diameter membranes using an Amicon 8010 stirred, dead-end filtration cell (Millipore) with a cell volume of 10 mL and an effective filtration area of 4.1 cm2 , attached to a 3.5-L dispensing vessel. Permeate was collected at fixed time

83

intervals using a FRAC-100 fraction collector (Pharmacia) and weighed to determine trans-membrane flux. Solute rejection was determined by UV spectroscopy. Filtration cells were stirred at 500 rpm using a stir plate to minimize concentration polarization. Deionized (DI) water was first passed through the membrane until the flux remained stable over at least a half hour (2 h minimum DI water filtration). The end of the stabilization period was taken to be the zero time point in the filtration plots. The cell was then emptied and refilled with the model foulant solution. Protein solutions comprised 1000 mg/L BSA in PBS. NOM fouling studies used 1000 mg/L humic acid and 10 mM CaCl2 in DI water. Polysaccharide fouling experiments were performed with 1000 mg/L sodium alginate in DI water. A sample of the permeate was collected after 1 h of filtration. Foulant retention values were obtained by measuring the foulant concentration in this sample by UV–vis spectroscopy using a Cary 500i UV–vis–NIR dual-beam spectrophotometer. The concentrations were quantified using UV absorbance at 280 nm for BSA and 200 nm for sodium alginate and humic acid. After each fouling test, the filtration cell was rinsed 5–7 times with DI water and then refilled with DI water as a feed to determine the reversibility of fouling. Flux decline experiments using activated sludge were performed following the same approach as described above, but with a different experimental setup. Filtration was performed using an Amicon stirred cell model 8200 (Millipore, Billerica, MA). The Amicon cell was mixed at 175 rpm for all sludge experiments. Flux measurements were performed by weighing permeate at fixed time intervals on a top loading balance (model PB3002-S, Mettler Toledo, Columbus, OH) assuming a permeate density of 1 g/mL. Automatic collection of permeate data was performed using the software WinWedge (TAL Technologies, Philadelphia, PA). Clean water flux and hydraulic recovery measurements were performed using NANOpure ultrapure water (Barnstead, Dubuque, IA). Activated sludge concentrations for dead-end filtration experiments using TFC or base membranes were approximately equal at 1745 and 1855 mg/L MLVSS, respectively. Membrane retention was characterized based on total COD measured in the filtrate and the retentate. The PVDF UF base membrane served as the control in filtration experiments, as UF membranes are commonly employed in MBRs. The fouling properties of commercial TFC NF membranes have been documented in other studies [18,30,31]. 2.5. Atomic force microscopy (AFM) measurements Atomic force microscopy (AFM) was used to characterize the interfacial forces between the membrane and organic foulants. The force measurements were performed with a MultiMode AFM connected to a Nanoscope IIIa controller (Veeco Metrology Group, Santa Barbara, CA). A carboxylate modified latex (CML) particle (Interfacial Dynamics Corp., Portland, OR) was used for making the AFM colloid probe because the model organic foulants used in this study (BSA, alginate and humic acid) contain predominantly carboxylic functional groups. The CML particle (3.0 ␮m in diameter) was attached by Norland

84

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

optical adhesive (Norland Products Inc., Cranbury, NJ) to a tipless SiN cantilever having a spring constant of 0.06 N/m (Veeco Metrology Group, Santa Barbara, CA) and cured under UV light for 20 min. A fluid cell was used to allow force measurements in desired solution chemistries, with the membrane being located on the bottom of the fluid cell. The fluid cell was first rinsed with DI water before injecting test solution to fully displace the DI water in the fluid cell. The force measurements were conducted at four different locations on the membrane, and 10 force measurements were taken at each location. Details on the procedures of using an AFM colloid probe to determine the intermolecular adhesion forces in membrane fouling are given in our recent work [30,32,33].

3. Results and discussion

Typical SEM micrographs of the base and coated membranes are shown in Fig. 1. As seen in Fig. 1a, the SEPRO PVDF base membrane exhibits a surface morphology of sparse, ∼0.5 ␮m circular pores characteristic of PVDF membranes fabricated by immersion precipitation [34]. The relatively high permeability of this membrane (pure water permeability of 2700 L/m2 h MPa) can be attributed to large macrovoids seen in cross-section (Fig. 1b). After coating, surface pores are no longer observable, and nodular features appear (Fig. 1c). From SEM cross-sections, the coating layer is approximately 2 ␮m thick (Fig. 1d). During fabrication PEG (Mn = 600 g/mol) was added to the coating solution and the coating was precipitated in ethylene glycol in order to swell the hydrophilic POEM domains and thereby improve the permeability of the membrane. Pure water permeabilities for the PVDF-g-POEM/PVDF membranes are consistent with those reported by Akthakul et al. [20], ranging between 20 and 56 L/m2 h MPa.

3.1. Membrane characterization 3.2. Fouling with model organic foulants The PVDF-g-POEM graft copolymers used in preparing the TFC NF membranes had a number-average molecular weight of 180 kg/mol, calculated from the initial molecular weight of the PVDF homopolymer and the weight percentage of POEM measured by NMR. The PEO content was 40 wt.% as determined from NMR spectroscopy. A control coater was used to achieve a uniform and reproducible PVDF-g-POEM coating thickness on the PVDF UF membrane substrates from concentrated DMF solution. Contact angle studies on PVDF-g-POEM coated membranes reported previously demonstrate that these coatings exhibit spontaneous wetting by water [20].

Previous studies on PVDF-g-POEM-coated membranes demonstrated complete resistance to fouling by a 1000 mg/L oleic acid/triethanolamine/water microemulsion over a 1.5 h filtration period, along with complete (>99.9%) retention of oleic acid [20]. Long-term fouling behavior, however, has much greater relevance to applications such as membrane bioreactor processes, as more fouling mechanisms can be taken into account [18,31,35,36]. Therefore, 10-day dead-end filtration studies were performed with the coated NF membranes, using foulants representative of proteins, polysaccharides and NOM,

Fig. 1. SEM micrographs of the PVDF400 UF base ((a) surface and (b) cross-section) and PVDF-g-POEM-coated ((c) surface and (d) cross-section) membranes.

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

along with filtration studies of shorter duration (16–24 h) with the PVDF400 UF base membrane control. In each case, Milli-Q water was passed through the membrane until the flux stabilized. The flux decline during this initial period is due to membrane compaction. With the application of pressure the base membrane is compressed, resulting in partial blocking of pores. In the data to follow, this effect is less notable for the NF membranes because the flux is largely determined by the non-porous coating layer, which does not deform substantially under pressure. The experimental setup included a small filtration cell (15 mL liquid capacity) connected to a large tank (3.5 L capacity) by a narrow tube. Therefore, during the filtration studies, diffusion from the cell to the tank was negligible. Since the coated NF membranes were observed to retain essentially all of the foulant, a constant build-up of foulant concentration occurred in the filtration cell, which could be estimated by a simple material balance around the cell. This aspect allowed characterization of the fouling resistance of the described membranes to very high foulant concentrations as well as the 10-day fouling potential in a single experiment. The continuous build up of foulant in these experiments further provides a close parallel to MBR operation, where high molecular weight substances can accumulate in the reactor over time. 3.2.1. Bovine serum albumin (BSA) Fig. 2a shows the 10-day dead-end filtration results from a TFC NF membrane for a 1000 mg/L solution of BSA in PBS, performed at 0.21 MPa (30 psi), and plotted as a function of normalized flux (flux/initial flux) versus time. The error bars in this figure (as well as in Figs. 3 and 4) arise from the limitations of the experimental setup, and not from collapsed data of

Fig. 2. Dead-end filtration of model protein solution with (a) PVDFg-POEM-coated NF membrane (average pure water permeability 39 ± 10 L/m2 h MPa) and (b) PVDF base membrane (average pure water permeability 2700 ± 660 L/m2 h MPa). () Milli-Q water; (䊉) 1000 mg/L BSA in PBS; (—) calculated concentration in the filtration cell (top graph only). Tests performed at 30 psi (a) and 10 psi (b).

85

Fig. 3. Dead-end filtration of model polysaccharide solution with (a) PVDF-gPOEM-coated NF membrane and (b) PVDF base membrane. () Milli-Q water; (䊉) 1000 mg/L sodium alginate in Milli-Q water; (—) calculated concentration in the filtration cell (top graph only). Tests performed at 30 psi (a) and 10 psi (b).

repeated experiments: the permeate flow comes one droplet at a time, and the measured volume at each data point can vary by the amount of a single droplet, approximately 0.05–0.1 mL, indicated by the error bars. For many data points, the error bars lie within the symbol area. The flux shows a slow decline over the course of the filtration (∼13% after 10 days), which is fully recovered when the cell is rinsed and the foulant solution is replaced with deionized water. Also shown in this figure is the calculated BSA concentration in the filtration cell, based on the

Fig. 4. Dead-end filtration of model NOM solution with (a) PVDF-g-POEMcoated NF membrane and (b) PVDF base membrane. () Milli-Q water; (䊉) 1000 mg/L humic acid in 10 mM CaCl2 ; (—) calculated concentration in the filtration cell (top graph only). Tests performed at 30 psi (a) and 10 psi (b).

86

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

measured BSA retention of >99.9% for this membrane (Table 1). The complete retention of BSA by the NF membrane is consistent with the reported globular dimensions of this protein (a heart-shaped molecule with 8 nm sides and 3 nm width [37]), relative to the width of the hydrophilic nanochannels of PVDFg-POEM observed by TEM (∼2 nm) [20,23]. At the end of 10 days, the cell concentration had reached 65,000 mg/L. The modest flux loss observed at such high concentrations suggests these membranes could be suitable in filtration operations involving highly concentrated feed streams. Fig. 2b shows the normalized flux data for the SEPRO PVDF400 UF base membrane for filtration of a 1000 mg/L BSA solution at 0.07 MPa (10 psi) over a period of 16 h. Despite the shorter run time, the irreversible flux loss for the UF membrane was around 60%. BSA retention for the UF base membrane after 1 h filtration was 69%, compared to complete retention (within instrument resolution) for the coated membrane (Table 1). The retention of the protein by the UF membrane is expected to increase in time due to internal pore fouling, as indicated by the large decline in flux. The results further illustrate the potential of these NF membranes to significantly enhance effluent quality. 3.2.2. Sodium alginate Similar results were obtained upon filtering 1000 mg/L sodium alginate, a model polysaccharide, through the coated and uncoated membranes (Fig. 3). In this case, the observed flux decline for both membranes was more dramatic. The TFC NF membrane showed approximately 71% loss relative to the pure water flux after 10 days of filtration at 0.21 MPa (30 psi). The coated membrane had an alginate retention of 92% after 1 h filtration, suggesting that low molecular weight portions of the foulant could pass through the membrane. The 92% figure can be considered a minimum retention for sodium alginate, since the substantial flux decline over the 10-day filtration was likely accompanied by increased retention. Assuming the retention remained constant throughout the 10-day experiment, Fig. 3a plots the calculated foulant concentration in the filtration cell as a function of time. By this conservative estimate, the sodium alginate concentration reached 12,000 mg/L by the end of the 10-day period. The flux decline observed in Fig. 3a can thus be attributed to the substantial rise in osmotic pressure due to alginate accumulation. However, the initial pure water flux could again be fully recovered for the NF membrane upon rinsing the cell and switching the feed to DI water. By contrast, a flux loss of 96% was observed for the UF base membrane after only 24 h filtration at 0.07 MPa (10 psi), with 41% irreversible flux loss (Fig. 3b). Alginate retention was also substantially lower (60%) for the uncoated membrane (Table 1).

3.2.3. Humic acid Membrane susceptibility to fouling by humic acid (HA) was also investigated. The fouling characteristics of humic acid are highly dependent on solution chemistry [30,31,38,39]. Membrane fouling by humic acid has been found to increase with the presence of divalent cations, particularly Ca2+ . Complexation of Ca2+ ions with the carboxyl groups of humic acid leads to intermolecular linkages that result in the formation of a compact cake layer that significantly reduces flux. Hence in these studies, 10 mM CaCl2 was added to the humic acid feed solution. After 10 days of filtration of 1000 mg/L HA solution at 0.21 MPa (30 psi), the TFC NF membrane showed no decrease in flux (Fig. 4a). At the end of this period, the concentration was calculated to be 24,000 mg/L, based on a measured HA retention of 99%. However, during the filtration, humic acid was observed to precipitate from solution, aided by the high concentration of Ca2+ in the feed [24]. Thus, the calculated concentration values in Fig. 4 indicate the extent of HA retention, rather than the dissolved concentration of humic acid in the cell. Upon opening the cell at the end of the fouling period, no cake layer was observed on the membrane. The remarkable constant flux demonstrated for the NF membrane stands in marked contrast to the results for filtration of the same HA solution through the UF base membrane. After 24 h filtration through the base membrane at 0.07 MPa (10 psi), an irreversible flux loss of 43% was found (Fig. 4b), with an HA retention of 37% (Table 1). 3.3. Fouling with MBR activated sludge The results from the combined studies above suggest that the PVDF-g-POEM-coated membranes should exhibit lower fouling and produce higher quality effluents in MBR operations compared with UF membranes conventionally employed. To further explore this application, dead-end filtration experiments were performed on uncoated and coated membranes using activated sludge taken from an aerobic bioreactor. Fig. 5 compares the normalized flux data for the TFC NF membrane and the PVDF400 base membrane. For the NF membrane (run at 0.28 MPa) no flux decline was observed through the experiment, i.e., no reversible or irreversible fouling occurred over the 16 h run period (Fig. 5a). The slight increase in flux throughout the experiment was most likely caused by a temperature increase and a corresponding decrease of the water viscosity. (Because of the low inherent flux, heat generated by the stir plate can raise the temperature in the filtration cell by 2 or 3 ◦ C over 16 h.) The lack of fouling by the activated sludge was confirmed with a second NF membrane sample. By contrast, the PVDF400 base membrane (run at 0.07 MPa) showed extensive fouling (Fig. 5b). A flux loss of 84% was reached after 4 h of activated sludge fil-

Table 1 Foulant retention values of PVDF-g-POEM coated NF membrane and PVDF base membrane Membrane

BSA retention (%)

Humic acid retention (%)

Sodium alginate retention (%)

Total COD retention (%)

PVDF-g-POEM coated NF SEPRO PVDF400 base

>99.9 69

99 37

92 60

99.5 98

Retentions measured after 1 h of foulant filtration.

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

87

Fig. 6. Decay lengths of the interaction forces between the AFM particle probe and the PVDF-g-POEM membrane as a function of ionic strength and composition. The error bars indicate standard deviations. Measurements were carried out at room temperature (23 ◦ C). NaCl (); CaCl2 + NaCl ().

Fig. 5. Dead-end filtration of MBR activated sludge with (a) PVDF-g-POEM coated NF membrane (1745 mg/L VSS) and (b) PVDF base membrane (1855 mg/L VSS). () NANOpure water; (䊉) activated sludge. Tests performed at 40 psi (a) and 10 psi (b). Data and error bars represent averages and standard deviations of five consecutive flux measurements.

position, with a decay length much greater than that predicted for electrostatic interaction (Fig. 6). For instance, in 100 mM solution, the Debye screening length, a characteristic length of electrostatic interactions, is about 1 nm whereas the measured decay length for the PVDF-g-POEM membrane was 13 nm. These observations suggest the presence of repulsive steric forces which prevent the adsorption of organic foulants to the membrane surface, consistent with the absence of irreversible fouling seen in dead-end filtration of the model organic foulants and activated sludge from the MBR. The adhesion forces, determined from the retraction (pull-off) curves, also support our fouling behavior data. The magnitude of the adhesion forces for the PVDF UF base membrane and the novel PVDF-g-POEM NF membrane are presented in Fig. 7 for

tration, with 57% irreversible flux loss. The NF membrane also exhibited better total COD retention: 99.5% compared with 98% for the base membrane (Table 1). 3.4. Interfacial forces and fouling resistance mechanism Interaction forces between the carboxylate modified latex particle probe (surrogate for organic foulants) and the membrane were determined at a range of ionic strengths (1–100 mM NaCl) as well as in the presence of 0.5 mM Ca2+ at a fixed ionic strength of 10 mM. Both approach (extending) and adhesion (retracting) force profiles were analyzed. The approach curves provide information on the type and the range of interaction forces between the foulant and the membrane, while the retracting (pull-off) curves provide information on the strength of foulant adhesion to the membrane surface. We have demonstrated previously [30,33] that the magnitude of the adhesion force correlates well with the fouling propensity of membranes in the presence of organic foulants. The interaction force profiles for the PVDF-g-POEM NF membrane showed little variation with ionic strength and com-

Fig. 7. Variation of the adhesion force normalized with the AFM particle probe radius as a function of solution ionic composition for the PVDF base UF membrane and the PVDF-g-POEM NF membrane. The experiments with divalent calcium ions were carried out with 0.5 mM CaCl2 plus 8.5 mM NaCl so that the total ionic strength was fixed at 10 mM. The error bars indicate standard deviations. Measurements were carried out at room temperature (23 ◦ C). PVDFg-POEM: NaCl (); CaCl2 + NaCl (). PVDF base: NaCl (); CaCl2 + NaCl (䊉).

88

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89

various solution ionic compositions. For the PVDF base membrane control, strong adhesion forces are detected as indicated by the large negative values of the measured adhesion force (F) normalized by the CML particle radius (R). Such strong adhesion forces correlate quite well with the severe organic fouling observed with this membrane. In contrast, no adhesion was detected with the PVDF-g-POEM-coated membrane (no negative values of F/R), even in the presence of divalent calcium ions. This remarkable observation agrees well with the lack of irreversible fouling observed for these membranes, even at high feed concentrations of organic foulants. 4. Conclusion Novel nanofiltration membranes prepared by coating a commercial PVDF UF membrane support with a self-assembling graft copolymer, PVDF-g-POEM, exhibited exceptional fouling resistance for a variety of model biofoulant solutions, including BSA, humic acid, sodium alginate, and activated sludge at feed concentrations of 1000 mg/L and above. The antifouling properties of these membranes can be attributed to both the nanoscale dimensions of the hydrophilic channels through the coating, which greatly restrict the size of permeate species, and to the unique properties of PEO, a charge-neutral polymer that hydrogen bonds with water to create an energetic barrier to the adsorption of biomolecules at the membrane surface. The fouling resistance of the novel NF membrane was verified by interfacial force measurements with AFM, showing the presence of steric foulant—membrane repulsive forces and lack of adhesion forces between the foulant and the membrane. Their combined anti-fouling character and high effluent quality suggest these membranes as promising prospective replacements for UF membranes in MBRs, where high concentrations of organic foulants are encountered. Although the pure water permeability of these NF membranes was much lower than that of UF membranes currently employed in MBRs, reducing the coating thickness should substantially improve flux performance. The PVDF-g-POEM coatings in this work are roughly 10× the thickness of polyamide selective layers on current commercial TFC NF membranes, suggesting a 10-fold enhancement in permeability should be possible. Considering the large decrease in UF membrane flux in the presence of foulants, the novel NF membranes described herein could potentially exhibit fluxes comparable to UF membranes under MBR operating conditions while resisting fouling and producing higher quality water. Acknowledgments The authors acknowledge Long Hua Lee for his assistance with the filtration studies, and Dr. Richard Salinaro of Pall Corp. for useful discussions. Financial support for this work was provided by the WaterCAMPWS, a Science and Technology Center of Advanced Materials for the Purification of Water with Systems under the National Science Foundation agreement number CTS-0120978, and by the U.S. Office of Naval Research under Award N00014-02-1-0343. This work made use of MRSEC

Shared Experimental Facilities supported by the National Science Foundation under Award DMR-0213282. References [1] I.S. Chang, P. Le Clech, B. Jefferson, S. Judd, Membrane fouling in membrane bioreactors for wastewater treatment, J. Environ. Eng. -Asce 128 (2002) 1018–1029. [2] J.A. Howell, T.C. Arnot, W. Liu, Membrane bioreactors for treating waste streams, Adv. Membr. Technol. 984 (2003) 411–419. [3] W.B. Yang, N. Cicek, J. Ilg, State-of-the-art of membrane bioreactors: worldwide research and commercial applications in North America, J. Membr. Sci. 270 (2006) 201–211. [4] N. Hilal, H. Al-Zoubi, N.A. Darwish, A.W. Mohammad, M. Abu Arabi, A comprehensive review of nanofiltration membranes: treatment, pretreatment, modelling and atomic force microscopy, Desalination 170 (2004) 281–308. [5] T.F. Speth, A.M. Gusses, R.S. Summers, Evaluation of nanofiltration pretreatments for flux control, Desalination 130 (2000) 31–44. [6] P.H. Hodgson, G.L. Leslie, R.P. Schneider, A.G. Fane, C.J.D. Fell, K.C. Marshall, Cake resistance and solute rejection in bacterial microfiltration—the role of the extracellular-matrix, J. Membr. Sci. 79 (1993) 35–53. [7] S. Rosenberger, H. Evenblij, S.T. Poele, T. Wintgens, C. Laabs, The importance of liquid phase analyses to understand fouling in membrane assisted activated sludge processes—six case studies of different European research groups, J. Membr. Sci. 263 (2005) 113–126. [8] H. Evenblij, J.H.J.M. van der Graaf, Occurrence of EPS in activated sludge from a membrane bioreactor treating municipal wastewater, Water Sci. Technol. 50 (12) (2004) 293–300. [9] N. Park, B. Kwon, I.S. Kim, J.W. Cho, Biofouling potential of various NF membranes with respect to bacteria and their soluble microbial products (SMP): characterizations, flux decline and transport parameters, J. Membr. Sci. 258 (2005) 43–54. [10] G.P. Sheng, H.Q. Yu, Characterization of extracellular polymeric substances of aerobic and anaerobic sludge using three-dimensional excitation and emission matrix fluorescence spectroscopy, Water Res. 40 (2006) 1233–1239. [11] B. Jin, B.M. Wilen, P. Lant, Impacts of morphological, physical and chemical properties of sludge flocs on dewaterability of activated sludge, Chem. Eng. J. 98 (2004) 115–126. [12] B. Frolund, R. Palmgren, K. Keiding, P.H. Nielsen, Extraction of extracellular polymers from activated sludge using a cation exchange resin, Water Res. 30 (1996) 1749–1758. [13] A. Bhattacharya, B.N. Misra, Grafting: a versatile means to modify polymers—techniques, factors and applications, Prog. Polym. Sci. 29 (2004) 767–814. [14] J.H. Choi, S. Dockko, K. Fukushi, K. Yamamoto, A novel application of a submerged nanofiltration membrane bioreactor (NF MBR) for wastewater treatment, Desalination 146 (2002) 413–420. [15] J.H. Choi, K. Fukushi, K. Yamamoto, Comparison of treatment efficiency of submerged nanofiltration membrane bioreactors using cellulose treacetate and polyamide membrane, Water Sci. Technol. 51 (6/7) (2005) 305– 312. [16] T. Wintgens, M. Gallenkemper, T. Melin, Endocrine disrupter removal from wastewater using membrane bioreactor and nanofiltration technology, Desalination 146 (2002) 387–391. [17] R.W. Baker, Membrane Technology and Applications, John Wiley, Chichester, New York, 2004. [18] M. Manttari, L. Puro, J. Nuortila-Jokinen, M. Nystrom, Fouling effects of polysaccharides and humic acid in nanofiltration, J. Membr. Sci. 165 (2000) 1–17. [19] R.J. Petersen, Composite reverse osmosis and nanofiltration membranes, J. Membr. Sci. 83 (1993) 81–150. [20] A. Akthakul, R.F. Salinaro, A.M. Mayes, Antifouling polymer membranes with subnanometer size selectivity, Macromolecules 37 (2004) 7663– 7668.

A. Asatekin et al. / Journal of Membrane Science 285 (2006) 81–89 [21] A. Akthakul, A.I. Hochbaum, F. Stellacci, A.M. Mayes, Size fractionation of metal nanoparticles by membrane filtration, Adv. Mater. 17 (2005) 532–535. [22] J.F. Hester, P. Banerjee, A.M. Mayes, Preparation of protein-resistant surfaces on poly(vinylidene fluoride) membranes via surface segregation, Macromolecules 32 (1999) 1643–1650. [23] J.F. Hester, P. Banerjee, Y.Y. Won, A. Akthakul, M.H. Acar, A.M. Mayes, ATRP of amphiphilic graft copolymers based on PVDF and their use as membrane additives, Macromolecules 35 (2002) 7652–7661. [24] A.R. Costa, M.N. De Pinho, M. Elimelech, Mechanisms of colloidal natural organic matter fouling in ultrafiltration, J. Membr. Sci. 281 (2006) 716–725. [25] G.J. Wilson, A. Pruden, M.T. Suidan, A.D. Venosa, Biodegradation kinetics of MTBE in laboratory batch and continuous flow reactors, J. Environ. Eng. -Asce 128 (2002) 824–829. [26] F.A. Holland, F.S. Chapman, Liquid Mixing and Processing in Stirred Tanks, Reinhold Pub. Corp., New York, 1966. [27] APHA, WEF and AWWA, Standard Methods for the Examination of Water and Wastewater, 20th ed., American Public Health Association, Washington, DC, 1998. [28] E.D. Rhine, G.K. Sims, R.L. Mulvaney, E.J. Pratt, Improving the Berthelot reaction for determining ammonium in soil extracts and water, Soil Sci. Soc. Am. J. 62 (1998) 473–480. [29] S. Inceoglu, S.C. Olugebefola, M.H. Acar, A.M. Mayes, Atom transfer radical polymerization using poly(vinylidene fluoride) as macroinitiator, Des. Monom. Polym. 7 (2004) 181–189. [30] Q.L. Li, M. Elimelech, Organic fouling and chemical cleaning of nanofiltration membranes: measurements and mechanisms, Environ. Sci. Technol. 38 (2004) 4683–4693.

89

[31] A.I. Schafer, A.G. Fane, T.D. Waite, Nanofiltration of natural organic matter: removal, fouling and the influence of multivalent ions, Desalination 118 (1998) 109–122. [32] W.S. Ang, S.Y. Lee, M. Elimelech, Chemical and physical aspects of cleaning of organic-fouled reverse osmosis membranes, J. Membr. Sci. 272 (2006) 198–210. [33] S. Lee, M. Elimelech, Relating organic fouling of reverse osmosis membranes to intermolecular adhesion forces, Environ. Sci. Technol. 40 (2006) 980–987. [34] A. Akthakul, W.F. McDonald, A.M. Mayes, Noncircular pores on the surface of asymmetric polymer membranes: evidence of pore formation via spinodal demixing, J. Membr. Sci. 208 (2002) 147–155. [35] Y. Li, L.-Y. Chu, J.-H. Zhu, H.-D. Wang, S.-L. Xia, W.-M. Chen, Thermoresponsive gating characteristics of poly(N-isopropylacrylamide)-grafted porous poly(vinylidene fluoride) membranes, Ind. Eng. Chem. Res. 43 (2004) 2643–2649. [36] A. Seidel, M. Elimelech, Coupling between chemical and physical interactions in natural organic matter (NOM) fouling of nanofiltration membranes: implications for fouling control, J. Membr. Sci. 203 (2002) 245–255. [37] M.L. Ferrer, R. Duchowicz, B. Carrasco, J.G. de la Torre, A.U. Acuna, The conformation of serum albumin in solution: a combined phosphorescence depolarization-hydrodynamic modeling study, Biophys. J. 80 (2001) 2422–2430. [38] W. Yuan, A.L. Zydney, Effects of solution environment on humic acid fouling during microfiltration, Desalination 122 (1999) 63–76. [39] S.H. Yoon, C.H. Lee, K.J. Kim, A.G. Fane, Effect of calcium ion on the fouling of nanofilter by humic acid in drinking water production, Water Res. 32 (1998) 2180–2186.